diff --git a/src/affine.rs b/src/affine.rs
index 7404293..b125389 100644
--- a/src/affine.rs
+++ b/src/affine.rs
@@ -1,3 +1,4 @@
+//! Manages the affine image transformations.
 use crate::Image;
 
 impl<const CHANNELS: usize> Image<Vec<u8>, CHANNELS> {
diff --git a/src/lib.rs b/src/lib.rs
index 59cd542..c45bde2 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,3 +1,6 @@
+//! # fimg
+//!
+//! Provides fast image operations, such as rotation, flipping, and overlaying.
 #![feature(
     slice_swap_unchecked,
     slice_as_chunks,
@@ -7,12 +10,13 @@
     test
 )]
 #![warn(
+    clippy::missing_docs_in_private_items,
     clippy::multiple_unsafe_ops_per_block,
     clippy::missing_const_for_fn,
     clippy::missing_safety_doc,
     unsafe_op_in_unsafe_fn,
     clippy::dbg_macro,
-    clippy::perf
+    missing_docs
 )]
 #![allow(clippy::zero_prefixed_literal)]
 
@@ -22,6 +26,10 @@ mod affine;
 mod overlay;
 pub use overlay::{Overlay, OverlayAt};
 
+/// like assert!(), but causes undefined behaviour at runtime when the condition is not met.
+///
+/// # Safety
+/// UB if condition is false.
 macro_rules! assert_unchecked {
     ($cond:expr) => {{
         if !$cond {
@@ -54,6 +62,7 @@ impl Image<&[u8], 3> {
     }
 }
 
+/// calculates a column major index, with unchecked math
 #[inline]
 unsafe fn really_unsafe_index(x: u32, y: u32, w: u32) -> usize {
     // y * w + x
@@ -98,7 +107,7 @@ impl<T, const CHANNELS: usize> Image<T, CHANNELS> {
     #[inline]
     /// create a new image
     pub const fn new(width: NonZeroU32, height: NonZeroU32, buffer: T) -> Self {
-        Image {
+        Self {
             buffer,
             width,
             height,
@@ -229,13 +238,15 @@ impl<const CHANNELS: usize> Image<Vec<u8>, CHANNELS> {
     /// if width || height == 0
     #[must_use]
     pub fn alloc(width: u32, height: u32) -> Self {
-        Image {
+        Self {
             width: width.try_into().unwrap(),
             height: height.try_into().unwrap(),
             buffer: vec![0; CHANNELS * width as usize * height as usize],
         }
     }
 }
+
+/// helper macro for defining the save() method.
 macro_rules! save {
     ($channels:literal == $clr:ident ($clrhuman:literal)) => {
         impl Image<Vec<u8>, $channels> {
diff --git a/src/overlay.rs b/src/overlay.rs
index f28d820..1a059e1 100644
--- a/src/overlay.rs
+++ b/src/overlay.rs
@@ -1,3 +1,4 @@
+//! Handles image overlay
 use super::{assert_unchecked, really_unsafe_index, Image};
 use std::simd::SimdInt;
 use std::simd::SimdPartialOrd;
@@ -22,11 +23,10 @@ pub trait Overlay<W> {
 }
 
 #[inline]
+/// SIMD accelerated rgba => rgb overlay.
+///
+/// See [blit](https://en.wikipedia.org/wiki/Bit_blit)
 unsafe fn blit(rgb: &mut [u8], rgba: &[u8]) {
-    const LAST4: Simd<u8, 16> = Simd::from_array([
-        255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 0, 0, 0, 0,
-    ]);
-
     let mut srci = 0;
     let mut dsti = 0;
     while dsti + 16 <= rgb.len() {
@@ -38,7 +38,9 @@ unsafe fn blit(rgb: &mut [u8], rgba: &[u8]) {
             threshold,
             [3, 3, 3, 7, 7, 7, 11, 11, 11, 15, 15, 15, 0, 0, 0, 0]
         );
-        mask &= LAST4;
+        mask &= Simd::from_array([
+            255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 0, 0, 0, 0,
+        ]);
 
         let new_rgb = simd_swizzle!(new, [0, 1, 2, 4, 5, 6, 8, 9, 10, 12, 13, 14, 0, 0, 0, 0]);
         let blended = (new_rgb & mask) | (old & !mask);
@@ -106,8 +108,16 @@ impl OverlayAt<Image<&[u8], 4>> for Image<&mut [u8], 3> {
 }
 
 impl OverlayAt<Image<&[u8], 3>> for Image<&mut [u8], 3> {
+    /// Overlay a RGB image(with) => self at coordinates x, y.
+    /// As this is a `RGBxRGB` operation, blending is unnecessary,
+    /// and this is simply a copy.
+    ///
+    /// # Safety
+    ///
+    /// UB if x, y is out of bounds
     #[inline]
     unsafe fn overlay_at(&mut self, with: &Image<&[u8], 3>, x: u32, y: u32) -> &mut Self {
+        /// helper macro for defining rgb=>rgb overlays. allows unrolling
         macro_rules! o3x3 {
             ($n:expr) => {{
                 for j in 0..($n as usize) {